The Potential for Bias across GPS-Accelerometer Combined Wear Criteria among Adolescents.

ABSTRACT

Physical activity has many health benefits, yet a large portion of our population is not meeting recommendations. Using accelerometry and global positioning systems (GPS) to accurately measure where people are active and to identify barriers and facilitators of activity across various settings will inform evidence-based policies and interventions to improve activity levels. Criteria for sufficient accelerometry data (e.g., number of days, minimum hours in a day) to accurately monitor free-living physical activity in adults and children have been widely studied, implemented, and reported by researchers. However, few best practice recommendations for researchers using GPS have been established. Therefore, this paper examined the impact of three co-wear criteria of varying stringency among a sample of children aged 10 to 16 years in Baton Rouge, Louisiana. Overall and location-based physical activity was consistent across the samples even within sociodemographic subgroups. Despite the lack of significant subgroup-specific mean differences in physical activity across the three samples, associations between sociodemographics and weight status and physical activity were significantly different depending on the device time-matching "co-wear" criteria applied. These differences demonstrate the critical impact co-wear criteria may have on conclusions drawn from research examining health disparities. There is a need for additional research and understanding of ideal co-wear criteria that reduce bias and accurately estimate free-living location-based physical activity across diverse populations.